Applied AI Engineer
I build LLM-powered systems that make knowledge easier to find and trust.
I’m an Applied AI Engineer at RICOH, working on the native workplace platform. I give AI tools better guidance so there’s less room to drift. My mission is to enable problem-solving for all of humanity by removing barriers to knowledge.
Loading the latest year…
Use the arrow keys to explore days. On smaller screens, scroll horizontally for earlier weeks. Press Escape to close the day tooltip.
GitHub activity is unavailable right now. View the profile instead. (opens in a new tab)
Repo-level RAG and enabling implementation teams
AI-assisted work kept missing unwritten team rules. I built repo-level RAG across six native repos to give development and review agents the right context across 100+ PRs. I also built a LangChain Excel-to-SVG converter that cut floorplan deployment costs for the implementation team.
Local inference for lower-cost video generation
Local open-weight model for video generation settings. It uses four seconds by default and six when confident with 79% precision on those calls. It replaced a costly 15 second shared quota default setting.
Onigami · Enemy behaviour and A* pathfinding
Published Unity turn-based strategy prototype for Windows, macOS and Linux. I wrote the enemy AI, building a behaviour tree for the enemy turn and A* for hexagonal navigation. It reached 2,000+ downloads and drew reviews from content creators.
